272 
Fishery Bulletin 11 6(3-4) 
1.0 
0.5 
(0 
I 
- 0.5 
-10 
- 1.5 
Month 
A May 
▼ June 
♦ July 
• August 
■ September 
+ October 
S. crameri’ 
S babcocki 
> 
Distance from shore • 
Ul 
'NPGO 
S. helvomaculatus' 
nP 
$ 0.0- 
cd 
pdo ; 
SSB SS it # SUMMER 
SPRING " " 1 
* 1 
Temperature ” • 
CM 
X T 
S. aleulianus 
II 
S pmmger 
CM 
? 
S. enlomelas' 
▼ 
Salinity 9 
S elongatus * 
- 1.5 
-10 
-0 5 0 0 
Axis 1 = 28.3% 
0.5 
10 
Figure 3 
Nonmetric multidimensional scaling ordinations showing axes 1 and 2 for 
larval rockfishes (Sebastes spp.) identified visually and environmental co¬ 
variates after collection off Washington and Oregon during 2005-2008. En¬ 
vironmental data include sea surface temperature (SST) and salinity at the 
sea surface (1 m; SSS) and at a depth of 30 m, Upwelling Index (UI), Pacific 
Decadal Oscillation (PDO) index, and North Pacific Gyre Oscillation (NPGO) 
index. Asterisks (*) denote species with coefficients of correlation >0.15. Each 
symbol represents the species composition of larvae collected in a single com¬ 
plete haul of a midwater trawl ordinated in species space (e.g. arranged so 
that similar hauls are near each other and dissimilar hauls are farther from 
each other). The arrows indicate the directionality and strength (by length 
of arrow) of physical correlations between hauls and physical variables mea¬ 
sured during the collection period. 
between temperature at 30 m and the NPGO (r=-0.66, 
P- 0.003). Plowever, because each variable represented 
a unique spatial scale or in situ habitat, they were all 
retained for subsequent analysis. 
Community analyses 
The NMS ordination of rockfishes identified by visual 
means (Figs. 3 and 4) contained 7 species caught in 
89 of 93 tows where we had visual identifications, and 
described 73.8% of the variation in the data set (axis 
1=28.3%, axis 2=24.8%, and axis 3=20.6%). The stress 
for the final solution was 17.0 (instability=0, 158 it¬ 
erations), considered fair but interpretable (Kruskal, 
1964). The largest variation in community structure 
was explained by seasonal differences along axis 1 
(Suppl. Table 1) (online only). Summer months (July 
through October, r=0.767) and warmer SST at 1 m 
(r=0.446) were positively correlated with axis 1 scores; 
spring months (May and June) were negatively corre¬ 
lated (r=-0.767). The species most likely to be pres¬ 
ent during warm, summer months was the rosethorn 
rockfish. During spring, we were more likely to find 
canary rockfish (S. pinniger), darkblotched rockfish, 
and widow rockfish. Axis 2 was positively associated 
with offshore stations (r=G.394) and negatively asso¬ 
ciated with warmer temperatures at 50 m, indicating 
that axis 2 differentiated between on-shelf and off-shelf 
